Postoperative Correction in Idiopathic Scoliosis: Which Preoperative Imaging Technique Is Most Predictive?

INTRODUCTION

Preoperative radiographic assessment of curve flexibility in patients with idiopathic scoliosis is important to determine Lenke classification, operative levels, and potential postoperative correction. However, no consensus exists regarding the optimal technique. We compared measurements from supine side bending (SB) and intraoperative traction radiographs under general anesthesia (TUGA) with actual postoperative correction followed for 1 year.

METHODS

We identified 235 patients with idiopathic scoliosis who underwent posterior spinal fusion with pedicle screw instrumentation between 2010 and 2018 who had preoperative and postoperative radiographs including standing posterior-anterior (PA) and lateral radiographs, preoperative SB radiographs, and TUGA radiographs. Curves were categorized into proximal thoracic, main thoracic/thoracolumbar (MT), and distal thoracolumbar/lumbar (TL/L) curves. Flexibility was calculated from SB and TUGA radiographs. Correction rates were calculated from 1 month and 1 year radiographs postoperatively. Bending radiographs that correlated significantly with postoperative correction with P<0.10 were eligible for inclusion. Preoperative demographics, etiology, deformity details, and surgical details were included in the multivariate models.

RESULTS

On univariate analysis, TUGA radiographs correlated with postoperative correction at 1 month and 1 year on MT curves (r=0.214, P=0.001; r=0.209, P=0.001) and TL/L curves (r=0.280, P<0.001; r=0.181, P=0.006). Supine SB radiographs did not correlate with postoperative correction on either MT or T/TL curves. On multivariate analysis, major curve TUGA radiographs were independently associated with postoperative MT curve correction at 1 month (beta: 0.158, 95% confidence interval: 0.035-0.280, P=0.012) and 1 year (beta: 0.195, 95% confidence interval: 0.049-0.340, P=0.009). MT curve SB radiographs were not associated with postoperative major curve correction at 1 month (P=0.088).

CONCLUSIONS

TUGA radiographs independently correlated with postoperative main thoracic and distal thoracolumbar/lumbar curve correction at 1 month and 1 year postoperatively. SB radiographs independently correlated only with TL/L curve correction at 1 year postoperatively. However, this correlation was not as strong as TUGA correction (beta of 0.280 vs. beta of 0.092). TUGA radiographs appear superior to SB radiographs at predicting curve correction after surgery.

LEVEL OF EVIDENCE: Level III.